Detection of Bird Nests on Power Line Patrol Using Single Shot Detector

Abstract

The presence of the bird nests on the electric power tower becomes a hazard to the safety and stability of the transmission line. In recent years, detecting the bird nests on the transmission line by using drones is being one of the essential missions of power inspection. The migration of image processing methods from computer vision to power image identification has increasingly becoming a trend. The detection method combining Single Shot Detector and HSV color space filter is proposed in this paper to identify the bird nests by making use of the image features with a large color span under different illumination angles. The fine-tuned Single Shot Detector network is trained and utilized to identify the bird nests and the detection result is clipped which called sub-images. Then the sub-images are filtered by the selector based on HSV color space, who contains none object of bird nests can be removed by the pixel percentage. The experimental results show that the proposed method can accurately detect the bird nests in the testing transmission line inspection images, and the accuracy can be up to 98.23%. Compared with other single traditional methods, the proposed bird nests detection method combining the deep learning and the HSV color space filter greatly enhances the detection accuracy.